Title: Savoy Accessibility – Drop-In PDF Accessibility for WordPress (WCAG 2.1 AA)
Author: Savoy AI
Published: मई 11, 2026
Last modified: सितम्बर 22, 2026
---
Search plugins
# Savoy Accessibility – Drop-In PDF Accessibility for WordPress (WCAG 2.1 AA)
[Savoy AI](https://profiles.wordpress.org/peteclark3/) लिखले बारे
[डाउनलोड](https://downloads.wordpress.org/plugin/savoy-accessible.1.0.84.zip)
* [Details](https://bho.wordpress.org/plugins/savoy-accessible/#description)
* [समीक्षा](https://bho.wordpress.org/plugins/savoy-accessible/#reviews)
* [Installation](https://bho.wordpress.org/plugins/savoy-accessible/#installation)
* [Development](https://bho.wordpress.org/plugins/savoy-accessible/#developers)
[मदद](https://wordpress.org/support/plugin/savoy-accessible/)
## बिबरन
Savoy Accessibility scans the PDFs on your WordPress site and uses the Savoy cloud
service to generate WCAG 2.1 AA-friendly HTML versions of each one, so screen readers
and other assistive technologies can read the content directly. It also includes
a set of on-demand audits and opt-in hardening toggles for common WCAG issues elsewhere
on your site.
The plugin is intended for sites that publish a lot of PDF content (schools, universities,
government agencies, nonprofits, and businesses).
#### Key features
* Automatic PDF detection across posts, pages, and the media library
* Conversion of PDF text content to semantic HTML using the Savoy service
* Automatic scanned PDF detection — text is extracted directly when a text layer
is present; vision AI OCR is used when the entire document is image-only (e.g.
a scanned page). Embedded images in any PDF are described separately via vision
AI for alt text.
* Automatic alt-text generation for images inside PDFs
* Every converted document is automatically audited with axe-core (the accessibility
engine used in Chrome DevTools), shown as Pass, Warning, or Needs Review
* Generated documents are stored as a custom post type so you can edit them in
Gutenberg
* Adjustable text size and a real high-contrast reading mode in the document viewer,
remembered per visitor
* Optional AI chat sidebar in the document viewer (off by default)
* Inline accessibility badge next to PDF links
* Optional automatic link rewriting so visitors land on the HTML version
* Automatic scanning every 4 hours to catch newly uploaded PDFs, plus an instant
check when a visitor opens a document that’s still converting
* Works with any theme
#### On-demand audits
* Image alt text audit
* Generic link text audit (WCAG 2.4.4) — flags “click here”, “read more”, bare
URLs
* Heading hierarchy audit (WCAG 1.3.1, 2.4.6) — flags multiple H1s and skipped
levels
* Tables-without-headers audit (WCAG 1.3.1)
* Empty or missing button labels audit (WCAG 4.1.2)
#### Opt-in hardening toggles
* Inject `` if missing
* Visible focus outlines for keyboard navigation
* Respect `prefers-reduced-motion`
* “Skip to main content” link
#### Software as a Service
This plugin is a client for the Savoy service at [savoy.chat](https://savoy.chat/accessible).
PDF processing, OCR, HTML generation, and AI chat all run on Savoy’s servers. The
plugin only communicates with Savoy when you provide an API key. See the Savoy Terms
of Service and Privacy Policy at [savoy.chat/terms](https://savoy.chat/terms) and
[savoy.chat/privacy](https://savoy.chat/privacy).
The free portion of the plugin (PDF detection and the audit/hardening tools) does
not require an API key and does not contact Savoy’s servers. PDF conversion and
link rewriting require a Savoy subscription.
#### How it works
1. Install and activate the plugin.
2. Go to Settings > Savoy Accessibility. Click “Rescan all content” to see how many
PDFs are on your site.
3. (Optional) Sign up at savoy.chat/accessible, paste in your API key, and click Save
and Verify.
4. Detected PDFs are sent to the Savoy service for conversion.
5. Each converted document is stored as an Accessible Document post in WordPress,
which you can edit.
6. Links on your site are optionally rewritten to point at the accessible HTML version.
#### What happens to my original PDFs?
The plugin never modifies your original files. It creates parallel HTML pages and
optionally points visitors to them. You can disable link rewriting at any time.
## Screenshots
[⌊Documents list showing converted PDFs with ready status⌉⌊Documents list showing
converted PDFs with ready status⌉[
Documents list showing converted PDFs with ready status
[⌊On-demand WCAG audits: image alt text, generic link text, heading hierarchy, tables
without headers, and button labels⌉⌊On-demand WCAG audits: image alt text, generic
link text, heading hierarchy, tables without headers, and button labels⌉[
On-demand WCAG audits: image alt text, generic link text, heading hierarchy, tables
without headers, and button labels
[⌊Accessible HTML version of a converted PDF, with PDF View and Accessible HTML
tabs⌉⌊Accessible HTML version of a converted PDF, with PDF View and Accessible HTML
tabs⌉[
Accessible HTML version of a converted PDF, with PDF View and Accessible HTML tabs
[⌊Accessible Documents post type in WordPress admin⌉⌊Accessible Documents post type
in WordPress admin⌉[
Accessible Documents post type in WordPress admin
## Installation
1. Install the plugin from the WordPress Plugin Directory, or upload the zip via Plugins
> Add New > Upload.
2. Activate the plugin.
3. Go to Settings > Savoy Accessibility.
4. Click “Rescan all content” to see how many PDFs on your site need accessibility
work. This step is free and works without an API key.
5. (Optional) Sign up at [savoy.chat/accessible](https://savoy.chat/accessible), get
an API key, paste it in, and click “Save and Verify”.
6. Once an API key is connected, the plugin sends detected PDFs to Savoy for conversion.
## FAQ
### Is this free?
Scanning, PDF detection, the WCAG audits, and the hardening toggles work without
any account and without contacting an external service. Conversion and link rewriting
require a Savoy subscription.
### Does this guarantee compliance with any specific regulation?
No. The plugin generates WCAG 2.1 AA-friendly HTML, which is the conformance level
cited by Section 508 and by the updated ADA Title II rule. Whether your site as
a whole conforms depends on your theme, your other content, and your authoring practices.
We strongly recommend a professional accessibility audit for any formal compliance
certification.
### Is this an overlay?
No. The plugin generates real HTML pages with semantic structure, alt text, and
reading order, stored as WordPress posts. It does not inject an overlay widget.
### What about scanned PDFs (image-only PDFs)?
The Savoy service uses vision-AI OCR for scanned PDFs.
### Can I edit the generated HTML?
Yes. Each converted document is a custom post type. You can edit it in Gutenberg
like any other page; revisions are preserved.
### How does pricing work?
See [savoy.chat/accessible](https://savoy.chat/accessible) for current pricing and
plan tiers.
### Does this slow down my site?
The plugin adds a lightweight content filter on the frontend. PDF conversion and
OCR run on Savoy’s servers, not yours.
### What if I cancel my subscription?
Already-converted documents remain in your WordPress install. New conversions stop.
### Can I use this for Section 508?
The HTML output is designed to meet WCAG 2.1 AA, which Section 508 references as
its conformance standard. As with any accessibility work, a professional audit is
recommended for formal compliance certification.
## समीक्षा
There are no reviews for this plugin.
## Contributors & Developers
“Savoy Accessibility – Drop-In PDF Accessibility for WordPress (WCAG 2.1 AA)” is
open source software. The following people have contributed to this plugin.
Contributors
* [ Savoy AI ](https://profiles.wordpress.org/peteclark3/)
[Translate “Savoy Accessibility – Drop-In PDF Accessibility for WordPress (WCAG 2.1 AA)” into your language.](https://translate.wordpress.org/projects/wp-plugins/savoy-accessible)
### Interested in development?
[Browse the code](https://plugins.trac.wordpress.org/browser/savoy-accessible/),
check out the [SVN repository](https://plugins.svn.wordpress.org/savoy-accessible/),
or subscribe to the [development log](https://plugins.trac.wordpress.org/log/savoy-accessible/)
by [RSS](https://plugins.trac.wordpress.org/log/savoy-accessible/?limit=100&mode=stop_on_copy&format=rss).
## Changelog
#### 1.0.84
* Fix: a document whose source PDF was confirmed gone could stay stuck permanently
failing to convert, since the check that removes it only ran during a content
scan, not a status sync. “Sync All Doc Statuses” now runs that check too, so
a document with a dead source link is cleaned up the next time statuses are synced,
not only when content is rescanned.
#### 1.0.83
* Fix: syncing a document (manually or automatically) no longer changes its Author.
A page’s Author is set blank only when the plugin first creates it, then follows
whoever last saves it by hand. Syncing and reprocessing leave it exactly as it
was; the Revisions panel still shows an automatic replacement as an automatic
change.
#### 1.0.82
* Fix: PDFs deleted from the media library stayed in the Documents list, and their
accessible pages stayed live. Deleting a PDF now removes its document from the
list, moves its accessible page to the trash (so it can be restored), and deletes
Savoy’s stored copy. PDFs removed some other way (over FTP or from your host’s
file manager) are cleaned up by the next scan. Nothing is remembered about a
deleted PDF: if the same file is added again it is converted as a brand-new document.
Only files inside your own uploads folder that are confirmed gone are removed(
never external links, and never files that still have a media-library entry,
such as offloaded cloud storage), at most 25 per scan, and a page that still
links to a deleted PDF does not bring it back.
#### 1.0.81
* Change: an Accessible Document’s Author now shows who last edited it. Whenever
someone saves a document in the editor, they become its Author (their name also
appears on that revision). Changes made by the plugin itself, such as syncing
or reprocessing a document, are treated as automatic and leave the Author blank,
so an administrator clicking Sync is never recorded as an editor. If someone
deliberately picks a different Author in the editor, that choice is kept.
#### 1.0.80
* New: Accessible Documents now have an Author field and full revision history
like regular posts. Each save records who made it and when, and any earlier version
can be restored from the editor’s Revisions panel. The original converted version
is saved before a document’s first edit, so a person’s very first change can
be rolled back too; a newer automatic conversion is recorded as an automatic
change (no user), separate from human edits. Up to 25 revisions are kept per
document unless your site sets its own limit.
#### 1.0.79
* Change: the automatic background sync now checks documents that are still converting
first, so a document that just finished reaches your site on the very next run
instead of waiting for the full sweep to come around to it. It also checks up
to 100 documents per request (was 25), so a full sweep of a large site takes
far fewer requests.
#### 1.0.78
* Change: the automatic background sync now checks every document, not just the
most recent few. It keeps a saved position and picks up exactly where the previous
run stopped (up to 150 documents or about 20 seconds per run, wrapping around
after the last one), so no document can be skipped no matter how many a site
has. It now also re-checks documents that hit your monthly limit or were temporarily
not found, and can no longer overlap with a run that is still going.
#### 1.0.77
* Fix: documents could stay stuck showing “processing” even though they had finished
converting. Status checks always asked about the same batch of already-finished
documents and never reached newer ones still waiting. “Sync All Doc Statuses”
now walks every document exactly once in order (ten at a time) and its “X of
Y checked” counter ends at your real document count. The background sync now
checks unfinished documents before finished ones, and no longer asks about more
documents per run than the server answers.
#### 1.0.76
* Change: “Sync All Doc Statuses”‘s progress counter now shows a real “X of Y documents
checked” that climbs toward and stops exactly at the site’s actual document count,
instead of a raw touch count that could run past the real total.
#### 1.0.75
* Fix: “Sync All Doc Statuses” could loop far past the site’s actual document count
and never stop on its own, since already-up-to-date documents kept getting endlessly
re-selected as “needing a check.” It now stops once the genuine backlog is actually
cleared, and no longer lets already-settled documents crowd out ones still waiting
on a real status update.
* Fix: a long accessible HTML document’s later content could render past the bottom
of its white content area, spilling onto the page background behind it, instead
of the content area growing to fit.
#### 1.0.74
* Fix: on a multisite network using a shared network-wide API key, an invalid key
was incorrectly recorded as a successfully-registered instant-update subscription
and never retried. A rejection is now correctly detected and retried on the next
scan.
* Change: the Documents section’s action buttons have been consolidated for clarity.“
Refresh Statuses” is removed (it made the exact same check as “Sync All Documents”,
just once instead of repeatedly). “Rescan all content” is renamed to “Scan for
New PDFs”, and “Sync All Documents” is renamed to “Sync All Doc Statuses” to
better describe what each does. “Check for Plugin Update” moved to the Settings
section, next to the plugin’s version number.
#### 1.0.73
* New: your site now registers itself to be notified the moment a document finishes
converting, instead of only finding out on the next scheduled sync (still every
30 minutes as a backup). No setup needed — this happens automatically.
#### 1.0.72
* Fix: visiting a document’s page could redirect forever (“too many redirects”)
if the background sync that runs on page-view hit a local data inconsistency.
The redirect now only happens after confirming that sync actually succeeded;
a failure leaves the current page showing instead of looping.
#### 1.0.71
* Fix: a document could end up permanently stuck showing “pending” even after it
finished converting on the server, if the local status update that’s supposed
to follow a successful report never landed. Both the background sync and “Refresh
Statuses” now also double-check pending documents against the server (a harmless
no-op for one that’s genuinely never been reported) instead of only ever checking
already-reported ones.
#### 1.0.70
* Change: up to 50 documents are now reported for conversion per sync cycle (was
10). A single shared worker made a bigger burst a real bottleneck before; that’s
no longer the case, so the limit now exists mainly to stay within reasonable
bounds rather than as a hard infrastructure ceiling.
#### 1.0.69
* Change: background sync (checking for new PDFs, pulling in finished conversions)
now runs every 30 minutes instead of every 4 hours. Existing sites pick up the
new schedule automatically on the next page load, no reactivation needed.
#### 1.0.68
* New: clicking Reprocess on a document you’ve edited directly in WordPress now
warns you first, since the new version will overwrite those edits once it’s ready.
Confirming authorizes that one overwrite; canceling leaves your edit untouched
and does not reprocess.
#### 1.0.67
* Fix: automatic and manual syncs could silently overwrite a document you’d edited
directly in WordPress (e.g. in the block editor) if a newer version was also
available from Savoy. We now detect when a document’s content has diverged from
what we last synced and skip the automatic overwrite, flagging it “Edited locally”
in your documents list instead. Use Reprocess if you want the newer version anyway.
#### 1.0.66
* Fix: a document reprocessed on the server after it was already synced to your
site (for example, a super-admin support action) never automatically reached
your WordPress site — the background sync only ever checked documents that hadn’t
been synced yet. It now also picks up documents whose accessible HTML changed
after the fact, same as manually clicking “Refresh Statuses.”
#### 1.0.65
* Fix: the redundant “view the original PDF” link on the placeholder page, removed
in 1.0.61, had come back in 1.0.62-1.0.64 due to a packaging mistake on our end.
Removed again, for real this time.
#### 1.0.64
* Fix: navigation menu PDF links always jumped to the Accessible HTML tab regardless
of the Default Tab setting added in 1.0.63 — that setting only reached content
links, not menu links. Both now respect it consistently.
#### 1.0.63
* New: “Default Tab” setting (Settings > Savoy Accessibility) controls which tab—
PDF View or Accessible HTML — a visitor lands on, independent of the PDF Link
Display Mode. Previously this was hardcoded per mode (Silent/Prominent/Replace
always jumped to the HTML tab, Icon always landed on PDF View). Defaults to PDF
View.
#### 1.0.62
* Fix: deactivating and reactivating the plugin left every accessible-document
link showing a 404, since WordPress’s cached rewrite rules never got refreshed
to include the document permalink structure again. Existing sites self-heal on
the next page load after updating; reactivating the plugin fresh flushes correctly
from now on.
#### 1.0.61
* Change: the “still converting” placeholder page no longer shows a separate “view
the original PDF” link — the PDF View tab already shows the same file and is
the default active tab.
#### 1.0.60
* New: visiting a document’s page while it’s still converting now checks its live
status right then, instead of waiting for the next background sync or a manual
Sync click. If it’s finished, the page updates in place immediately.
#### 1.0.59
* Change: background sync (checking for new PDFs, pulling in finished conversions)
now runs every 4 hours instead of once a day. Existing sites pick up the new
schedule automatically on the next page load, no reactivation needed.
#### 1.0.58
* New: a document still converting now gets a real placeholder page on your own
site the moment it’s detected, instead of sending visitors to savoy.chat while
it processes — the same page updates in place with the finished accessible version
once conversion completes, so the URL never changes.
* New: “Sync All Documents” button pulls in every document that finished converting
server-side but hasn’t synced to this site yet, in one click (previously required
repeatedly clicking “Refresh Statuses”).
* New: “Reprocess All Documents” button re-runs the full conversion pipeline on
every document, with a warning since this uses real monthly quota.
* Fix: a server-side error (an expired subscription, a disabled account, an invalid
key) always showed a generic “Failed to…” message with no explanation, hiding
the real reason. Now shows the actual server response.
* Fix: silent-mode badge suppression, verified across every code path in the plugin,
not just the one case previously fixed.
#### 1.0.57
* Fix: a PDF link rewritten while its accessible version was still processing always
showed the accessibility badge icon, ignoring the “Silent rewrite” display mode
setting — the mode check only applied once a document was ready. Now respected
in both states.
#### 1.0.56
* New: “Clean Document Titles” setting (off by default) drops the “.pdf” file extension
from an accessible document’s page title and URL — “Employee-Handbook.pdf” becomes
a page titled “Employee-Handbook” at a matching URL, instead of literally including“.
pdf” in both.
#### 1.0.55
* Fix: PDF links placed in a widget (a sidebar Text widget, a footer Custom HTML
block, or any block added via the Widgets screen) were never detected or rewritten
to the accessible version, since widget content is stored separately from post/
page content. Now scanned and rewritten the same way.
#### 1.0.54
* Change: trimmed the Network Settings page copy.
#### 1.0.53
* Fix: corrected the Network Settings page copy, which incorrectly said Network
Activate was required for a child site to inherit the network-wide key. Normal
per-site activation is enough.
#### 1.0.52
* New: on WordPress Multisite, network admins can set one API key (Network Admin
> Savoy Accessible) that every site on the network uses automatically, no need
to activate the plugin or enter a key on each individual site. A site can still
connect its own separate key to override the network one just for itself.
* Fix: documents are now matched by their full URL instead of just their filename,
so two different sites sharing one account (e.g. a multisite network) can each
have a “handbook.pdf” without them being confused for the same document.
#### 1.0.51
* Fix: the WCAG validation check was testing the generated HTML in isolation, without
the same landmark both real viewer pages already wrap it in, causing a false “
Warnings” score on nearly every document. Fixed so the check matches what visitors
actually see.
* New: hover over a “Warnings” badge in the Documents list to see exactly what
was flagged.
* Improvement: documents now always get exactly one heading (title), even short
single-page letters that previously might have been generated without one.
#### 1.0.50
* Change: documents list defaults to 10 per page (was 25), and now shows “Showing
docs X – Y of Z” instead of “page X of Y”.
* New: a Per Page dropdown (10/25/50/100) lets you choose how many documents to
see at once.
#### 1.0.49
* New: Documents list is now paginated (25 per page, Previous/Next), instead of
silently cutting off after 100 documents. Sites with hundreds or thousands of
PDFs can now see and manage all of them.
* New: “Retry Failed Documents” button, retries up to 10 failed documents at a
time (click again for more) instead of retrying one at a time.
* Change: renamed “Check for Update” to “Check for Plugin Update” for clarity.
#### 1.0.48
* Fix: a crash in the backend conversion pipeline was causing “Processing failed:
cannot access local variable ‘Site’…” on every document for some sites. Backend-
only bug, fixed server-side.
* Change: “Refresh Statuses” no longer reports newly-scanned PDFs for conversion
as a side effect, it only checks the status of documents already submitted. Reporting
new PDFs stays with “Scan Now,” where it belongs.
#### 1.0.47
* Fix: your own site’s domain could get silently dropped from the PDF conversion
allowlist if you clicked Save Settings before ever filling in Allowed Domains,
blocking all PDF conversion until you noticed and re-added it by hand. Your own
connected domain is now always allowed regardless of what’s in that field.
* Change: removed the automatic status-check polling loop entirely. Statuses now
only refresh once when you load this page, or when you click “Refresh Statuses”;
no more background checking while the tab sits open.
#### 1.0.46
* Fix: a site with a large backlog of PDFs (100+) could bog down under the status-
check polling. The admin page now checks statuses in small batches (a single
combined request instead of one per document) once a minute instead of every
3 seconds, and only reloads the page when a status actually changes. Newly discovered
PDFs are now also reported for conversion in small batches instead of all at
once.
#### 1.0.45
* Added a tooltip on the “X / Y documents this month” usage count explaining that
documents over 30 pages count as more than one conversion, so the number can
run ahead of your actual document count.
#### 1.0.44
* New: reading preferences in the document viewer — adjustable text size and a
real high-contrast mode (pure black-on-white, not a color filter), both applied
to the viewer’s own semantic markup and remembered per visitor.
#### 1.0.43
* Change: rebranded from “Savoy AI” to “Savoy” throughout (plugin author field
only — no functional changes).
#### 1.0.42
* Change: the PDF domain allowlist now defaults to allowing only this site’s own
domain, instead of allowing every domain. Prevents the plugin from scanning and
converting PDFs linked from unrelated sites. Your own domain is always allowed
automatically; add others under Settings if you need to.
* Fix: the domain allowlist you configure is now synced to and enforced by the
Savoy server too, not just used locally for link rewriting.
#### 1.0.41
* New: PDF domain allowlist — restrict link rewriting and scanning to specific
domains. Prevents 403/404 errors from off-site PDF links.
#### 1.0.40
* New: “Silent rewrite” link mode — rewrites PDF links to the accessible HTML viewer
with no visual change to link text or styling.
#### 1.0.39
* Fix: database schema migrations now run automatically on plugin upgrade, not
just on fresh activation. Fixes cases where columns added in recent releases (
validation_score, etc.) were missing after an auto-update, which prevented processing
progress from being stored and displayed.
#### 1.0.38
* New: WCAG validation score shown alongside each converted document (Pass / Warnings/
Needs review), powered by axe-core.
#### 1.0.37
* Security: viewer URLs no longer contain the API key. The site ID (a read-only
identifier) is used instead, so API keys are never exposed in browser history
or server logs.
#### 1.0.36
* PDF links in navigation menus are now rewritten to point at the accessible HTML
version, the same as links in post content.
#### 1.0.35
* When HTML is regenerated server-side, Refresh Statuses now detects the change
and automatically re-syncs the accessible HTML to the WordPress post.
#### 1.0.34
* Error messages from document processing (e.g. broken PDF links, access errors)
now persist in the database and display below the status badge in the documents
list. Previously errors were only visible while the processing job was active.
#### 1.0.33
* Renamed plugin display name to Savoy Accessibility. No behavior changes.
#### 1.0.32
* Moved the three opt-in hardening styles (focus outline, reduced-motion, skip
link) out of inline `